Keep your agents useful without making risky actions invisible.

Safety Gate by ObfuscAIte adds policy, approvals, and auditability to MCP tool execution for local AI agents — especially shell and file access in OpenClaw-style workflows.

Approval-aware MCP security middleware for local-first agent workflows.
View on GitHub Read the Docs

Your agents are either too powerful, too restricted, or too opaque

⚖️

Direct shell and file access make risky actions hard to review

Once an agent can run commands or modify files directly, important changes can happen faster than humans can follow them.

👁️

"Just trust the agent" is not a security model

You need a middle layer that can allow safe actions, pause risky ones, and make the decision path reviewable.

🚫

Approvals and audits are usually bolted on too late

By the time someone asks who approved a risky action or whether it can be replayed, the system often has no clean answer.

A reviewable control layer for MCP tool execution

Safety Gate sits between local AI agents and risky shell/file actions, adding policy checks, approvals, and auditability.

❌ Without Safety Gate

Agent requests: write package.json + run shell command
🚨 Tool executes with little review or auditability
Unclear accountability | Risky changes | Opaque execution

✅ With Safety Gate

Agent requests: write package.json + run shell command
🛡️ Safety Gate evaluates policy and may require review
Human inspects the request, unified diff, and approval details
Execution is attributable, authenticated, expirable, and auditable

Policy-enforced shell and file execution

Allow low-risk actions automatically, deny unsafe patterns, and route higher-risk operations into review.

🌐

Built for MCP and OpenClaw-style local agent workflows

Designed for agent environments where local shell and file tools are powerful, but too important to leave invisible.

🔌

Approvals, expiry, replay protection, and audit trail

Know who approved, who executed, what changed, and whether a stale approval could be reused later.

Why Safety Gate vs. adjacent alternatives?

Most tools optimize for orchestration or isolation. Safety Gate focuses on reviewable MCP tool execution.

Capability Safety Gate Orchestrators Sandbox Runtimes Cloud Guardrails DIY Scripts
Policy-enforced tool execution ✓ Core product ✗ Usually not the focus ✗ Usually lower-level isolation ✗ Often provider-specific ✗ Hand-rolled
Human review for risky actions ✓ Built in △ Sometimes ✗ Usually not the focus ✗ Rarely ✗ Manual only
Approval auth + expiry + replay protection ✓ Built in ✗ Not typical ✗ Not typical ✗ Not the main model ✗ Custom work
Unified diff review for file writes ✓ Built in ✗ Not typical ✗ Not typical ✗ No ✗ DIY
OpenClaw / MCP-friendly ✓ Yes Varies Varies Varies ✓ If you build it
Positioning Approval-aware MCP middleware Workflow coordination Execution isolation Provider-layer guardrails Custom glue

Built by AI security experts

Proven track record in AI safety and vulnerability research

🏆

HackAPrompt #21 Globally

Ranked among top AI security researchers in global prompt injection competition

🔍

8 Disclosed Vulnerabilities

Responsible disclosure of critical security flaws through bug bounty programs

🏗️

Production RAG Experience

Built and scaled AI systems handling millions of sensitive queries in production

Add a reviewable control layer to your agent tools

If you want shell and file access without blind trust, Safety Gate is the layer in the middle.

View on GitHub Read the Docs 📅 Book a strategy chat
>